Subject: Re: killing the "l" argument from vfs
To: Antti Kantee <firstname.lastname@example.org>
From: Andrew Doran <email@example.com>
Date: 11/25/2007 18:34:19
On Sat, Nov 24, 2007 at 10:08:35PM +0200, Antti Kantee wrote:
> I'm working on a patch which will rid the vfs interfaces of the randomly
> teleporting "l". At least VFS_ and VOP_ will get the treatment, I still
> need to investigate componentname. This will clarify the interfaces,
> as currently the passed value for lwp might be NULL in some cases.
> The new way is to fetch curlwp if it's desired (which is already done
> in plenty of places due to "l" not being always passed)..
> If someone has problems with this, please scream very loudly very soon.
The last couple of times I looked, I didn't find a place in VFS where the
pointer being passed around was different from curlwp.
I can only guess that the whole 'curproc as an argument' thing was some
aincent tip of the hat to multi CPU support. I really don't know. We should
work towards getting rid of it in most places because it's confusing, people
end up coding for situations that don't occur, and it leads to bugs.